import numpy as np
import pandas as pd
import geopandas as gpd
import fiona
import glob
import os
import contextily as ctx
from scipy.spatial import cKDTree
from shapely.geometry import Point, shape, LineString, MultiLineString, GeometryCollection, MultiPoint, Polygon, MultiPolygon  # creating points
import json
from tqdm.auto import tqdm
from tqdm.contrib.concurrent import process_map, thread_map
pd.set_option('min_rows', 30)
import sys
from importlib import reload
from util import *
import matplotlib.pyplot as plt
import multiprocessing
# from pandarallel import pandarallel # parallel operations for speed
# pandarallel.initialize(nb_workers=8, progress_bar=True)
# import swifter
tqdm.pandas()
plt.rcParams['figure.figsize'] = (16, 16)
pd.set_option('display.max_columns', None)
pd.set_option('display.max_rows', 10)
gpd.options.use_pygeos 

True

ls('restricted')

	name	filesize (MB)	last modified
0	2016_Unitary_Plan_operational_15Nov	0.00	2021-09-02 04:46:20.562659
1	BCs_issued_by_AUP_TLADCs_2021FEB.csv	64.34	2021-09-23 04:36:31.125500

Total: 64.0MB

Setup

%%time
parcels = gpd.read_file('input/lds-nz-primary-parcels-FGDB.zip!nz-primary-parcels.gdb')
parcels = parcels[parcels.land_district.isin(['North Auckland', 'South Auckland'])].to_crs(4326)
parcels['geometry_polygon_4326'] = parcels.geometry
parcels['geometry_polygon_2193'] = parcels.geometry.to_crs(2193)

parcels['geometry_centroid_2193'] = parcels['geometry_polygon_2193'].centroid
parcels['representative_point_2193'] = parcels['geometry_polygon_2193'].representative_point()

parcels['geometry_centroid_4326'] = parcels['geometry_centroid_2193'].to_crs(4326)
parcels['representative_point_4326'] = parcels['representative_point_2193'].to_crs(4326)

CPU times: user 1min 23s, sys: 3min, total: 4min 24s
Wall time: 5min 39s

parcels_sample = parcels.sample(10000)

# keep a centroid and polygon version - change between them according to need
parcels_sample['geometry_polygon_2193'] = parcels_sample.geometry.to_crs(2193)
parcels_sample['geometry_centroid_2193'] = parcels_sample.geometry_polygon_2193.centroid
parcels_sample['geometry_centroid_4326'] = parcels_sample.geometry_centroid_2193.to_crs(4326)
parcels_sample['geometry_polygon_4326'] = parcels_sample.geometry_polygon_2193.to_crs(4326)

# bounds of the auckland region, including all parcels in dataset
bounds2193 = {'x1': 1.5e6, 'x2':2e6, 'y1':5.8e6, 'y2':6.1e6}

# get a sample of non-road polygons
parcels_sample2 = parcels_sample[parcels_sample.parcel_intent != 'Road'].sample(5)

parcels.sindex
parcels_sample.sindex
parcels_sample2.sindex

<geopandas.sindex.PyGEOSSTRTreeIndex at 0x7fda5c6eabe0>

1. LINZ parcel information

a. LINZ_parcel_ID Parcel identifier - “certificate of title”.

parcels['LINZ_parcel_ID'] = parcels.id

b. LINZ_parcel_centroid_lon

c. LINZ_parcel_centroid_lat

parcels['geometry'] = parcels.geometry_centroid_4326
parcels = parcels.set_crs(4326)
parcels['LINZ_parcel_centroid_lon'] = parcels.geometry.x
parcels['LINZ_parcel_centroid_lat'] = parcels.geometry.y

d. LINZ_parcel_vertices_lon vector of longitudes of the vertices of the matched LINZ parcels

e. LINZ_parcel_vertices_lat

%%time
def extract_verts(geometry):
    lat = np.nan
    lng = np.nan
    if geometry:
        coordinates = []
        for polygon in geometry:
            # the last point is the same as the first
            coordinates.extend(polygon.exterior.coords[:-1])
        lng = f"[{'; '.join([str(round(point[0], 6)) for point in coordinates])}]"
        lat = f"[{'; '.join([str(round(point[1], 6)) for point in coordinates])}]"
    return lng, lat

vertices = process_map(extract_verts, parcels.geometry, max_workers=14, chunksize=1000)
parcels["LINZ_parcel_vertices_lon"] = [v[0] for v in vertices]
parcels["LINZ_parcel_vertices_lat"] = [v[1] for v in vertices]

CPU times: user 1min 3s, sys: 26.7 s, total: 1min 30s
Wall time: 1min 12s

f. LINZ_parcel_roadvertices_lon subvector of longitudes of parcel that sits adjacent to a road

g. LINZ_parcel_roadvertices_lat

parcels['geometry'] = parcels.geometry_polygon_4326
parcels = parcels.set_crs(4326)

roads = parcels[parcels.parcel_intent == "Road"]
roads = roads.to_crs(parcels.crs)

%%time
roads_dissolved = roads.dissolve()

CPU times: user 26.8 s, sys: 724 ms, total: 27.5 s
Wall time: 27.5 s

def pointarray2matarrays(pointarray):
    """list of points to matlab compatible arrays of longs and lats
    
    i.e.
    [point1, point2,...] -> 'point1_x; point2_x; ...', 'point1_y; point2_y; ...'
    """
    lon = [point.xy[0][0] for point in pointarray]
    lat = [point.xy[1][0] for point in pointarray]
    lon = f"[{'; '.join([str(round(lon, 6)) for lon in list(lon)])}]"
    lat = f"[{'; '.join([str(round(lat, 6)) for lat in list(lat)])}]"
    return lon, lat

def get_points_in_roads(row):
    """return a list of points from the geometry that fall within roads_dissolved"""
    # assume multipolygon
    if row[1].parcel_intent == 'road':
        return None
    geom = row[1].geometry
    assert isinstance(geom, MultiPolygon), f"not implemented for geometry of type {type(geom)}"
    road_points = []
    # iterate over polygons
    for poly in geom:
        # split polygon into vertices
        # the last point is the same as the first
        coords = poly.exterior.coords[:-1]
        pointsx = [x for x, _ in coords]
        pointsy = [y for _, y in coords]
        # create gdf with one row per vertex
        points_gdf = gpd.GeoDataFrame(geometry=gpd.points_from_xy(pointsx, pointsy)).set_crs(4326)
        # sjoin with roads, to eliminate vertices that don't intersect a road
        road_points.extend(gpd.sjoin(points_gdf, roads_dissolved, op = 'intersects').geometry.values)
    # split into matlab compatible longs and lats like [(longs_list, lats_list), (longs_list, lats_list),...]
    road_points = pointarray2matarrays(road_points)
    return road_points

# this might hang for a few minutes before multiprocessing starts
road_intersections = process_map(get_points_in_roads, parcels.iterrows(), max_workers=14, chunksize=1, total=len(parcels))

parcels['LINZ_parcel_roadvertices_lon'] = [r[0] for r in road_intersections]
parcels['LINZ_parcel_roadvertices_lat'] = [r[1] for r in road_intersections]

# example
sample = parcels.iloc[[2065]]
road_points = get_points_in_roads(sample.iloc[0].geometry)
print(road_points)
ax = gpd.GeoDataFrame(geometry=road_points).plot()
sample.plot(ax=ax, alpha=0.5)
roads_dissolved.plot(ax=ax, color='red', alpha=0.5)
x1, y1, x2, y2 = sample.buffer(0.005).total_bounds
plt.xlim(x1, x2)
plt.ylim(y1, y2)
plt.show()

h. LINZ_adjoining_parcel_ID id of adjoining LINZ parcels

i. LINZ_parcel_sides_zones AUP Zone Code of adjoining parcels (this includes residential, business, and rural zones; it should also include roads, water and open spaces)

For convenience/speed, 1h and 1i can be gotten simultaneously.
Note: Do 2a and 2b here first, which are needed for 1i.

What to use for joining parcels with AUP zones?

• intersect: fails because a parcel's exterior might intersect with an AUP zone's exterior, without any overlap
• overlap: fails because it is false if one polygon is contained completely within the other
  To use polygon/polygon operations, a combination would be needed; something like intersects but does not touch.
  Better to use a representative point - which may miss cases when a parcel is spread over two AUP zones (not sure if that occurs).
  Bonus: representative point + within is many, many, many times faster than using intersects + not touches (~10s vs not finishing before killing after at least 30 minutes)

first, 2a and 2b

%%time
aup_zones = gpd.read_file('restricted/2016_Unitary_Plan_operational_15Nov/UP_OIP_15Nov2016_SHP/MASTER_UP_BaseZone.shp')
aup_zones.sindex
aup_zones.sample(3)

CPU times: user 25.8 s, sys: 2.4 s, total: 28.2 s
Wall time: 30.1 s

	WORKFLOWJO	last_edi00	CONTOUR	created_da	DocumentUR	GlobalID	GROUPZONE	GROUPZONE_	ID	NAME	OBJECTID	PARCEL_BAS	PARCEL_B00	PRECINCT	PRECINCT_r	PRECINCTGR	PRECINCT00	SCHEDULE	SHAPE_Area	SHAPE_Leng	STATUS	SUBPRECINC	SUBPRECI00	SUBTYPE	SUBTYPE_re	TYPE	TYPE_resol	VALIDATION	VALIDATI00	VERSIONSTA	VERSIONS00	ZONE	ZONE_resol	ZONEHEIGHT	geometry
4323	57767.0	20161111010	None	20160718211	None	{2653E2A1-4629-48CA-AADC-2038EFCAC51E}	2	Residential	None	Everglade School	4324	None	None	None	None	None	None	None	25532.898535	745.711232	None	None	None	None	None	None	None	3	Valid and Public	4	Operative	18	Residential - Mixed Housing Suburban Zone	NaN	POLYGON Z ((1768860.282 5903754.262 0.000, 176...
111639	NaN	20161111012	None	20160718211	None	{2AFF46AC-8E35-48EC-89C3-D0E18F776D78}	7	General	None	None	111640	None	None	None	None	None	None	None	3452.179432	398.432755	None	None	None	None	None	None	None	3	Valid and Public	4	Operative	27	Road	NaN	POLYGON Z ((1757930.155 5916271.915 0.000, 175...
47139	NaN	20161111011	None	20160718211	None	{1F572D18-2901-44B3-B7FA-5B9F8F71C99A}	5	Coastal	None	None	47140	None	None	None	None	None	None	None	0.835644	5.790642	None	None	None	None	None	None	None	3	Valid and Public	4	Operative	59	Coastal - Coastal Transition Zone	NaN	POLYGON Z ((1750889.741 5924939.607 0.000, 175...

# use 2193 for AUP; this will be useful later when calculating haversine distances to nearest zones
# representative point is much faster for sjoin
parcels['geometry'] = parcels.representative_point_2193
parcels = parcels.set_crs(2193)
aup_zones = aup_zones.to_crs(parcels.crs)

%%time
parcels_zoned = gpd.sjoin(parcels, aup_zones[['ZONE', 'ZONE_resol', 'geometry']], op='within').drop(columns=['index_right'])

CPU times: user 9.4 s, sys: 247 ms, total: 9.65 s
Wall time: 9.67 s

# check if there was a one to one match
print(len(parcels_zoned.index.unique()))
print(len(parcels_zoned))
print(len(parcels.index.unique()))
print(len(parcels))

537095
537118
537290
537290

# 23 parcels fall have their representative point fall under multiple AUP zone polygons
# but see below, only one of these falls in two AUP zone polygons with different zone codes
np.unique(parcels_zoned.index.value_counts(), return_counts=True)

(array([1, 2]), array([537072,     23]))

%%time
def get_parcel_zones(id):
    if id in parcels_zoned.index:
        zoned = parcels_zoned.loc[[id]]
        # filter out duplicates
        return ','.join(zoned.ZONE.unique())
    else:
        return None
parcel_zones = process_map(get_parcel_zones, list(parcels.index), max_workers=14, chunksize=1000)

CPU times: user 2.32 s, sys: 1.49 s, total: 3.81 s
Wall time: 43min 3s

# there is a one to one relationship between zones and zone names
print(len(aup_zones[['ZONE', 'ZONE_resol']].drop_duplicates().ZONE.unique()))
print(len(aup_zones[['ZONE', 'ZONE_resol']].drop_duplicates()))
zone2zone_name = {r.ZONE: r.ZONE_resol for _, r in aup_zones[['ZONE', 'ZONE_resol']].drop_duplicates().iterrows()}

50
50

# only one parcel has its representative point fall in two zones
print([z for z in parcel_zones if z is not None and ',' in z])
print(zone2zone_name['18'])
print(zone2zone_name['19'])

['18,19']
Residential - Mixed Housing Suburban Zone
Residential - Single House Zone

# take the first zone where there are multiple (only one case)
parcels['LINZmatch_AUP_code'] = [z.split(',')[0] if (z is not None) else None for z in parcel_zones]
parcels['LINZmatch_AUP_name'] = [zone2zone_name[z.split(',')[0]] if (z is not None) else None for z in parcel_zones]

now do 1h & 1i (now that parcels have zones)

# need polygons, will check for touching neighbours
parcels.geometry = parcels['geometry_polygon_4326']
parcels = parcels.set_crs(4326)

%%time
# number of rows to process at once
n_rows=100
def find_neighbour_info(i):
    """find neighbours of parcels from i * n_rows to (i+1) * n_rows, then find ids and zones of those neighbouring parcels"""
    neighbour_gdf = gpd.sjoin(parcels.iloc[i*n_rows:min((i+1)*n_rows, len(parcels)-1)], parcels, op='touches')
    neighbour_zones = []
    for idx in neighbour_gdf.index:
        neighbour_ids.append(neighbour_gdf.loc[[idx]].id_right.tolist())
        neighbour_zones.append(neighbour_gdf.loc[[idx]].LINZmatch_AUP_code_right.tolist())
    return neighbour_ids, neighbour_zones
        
# each call to find_neighbours will return two lists like this: [list of neighbour ids], [list of niehgbour zones]
# The final structure will be of shape (n_chunks, 2, n_neighbours), where n_neighbours will vary between lists
parcel_neighbour_chunks = process_map(find_neighbour_info, list(range((int(np.ceil(len(parcels) / n_rows))))), max_workers=14, chunksize=10)

parcel_neighbour_chunks

parcels['LINZ_adjoining_parcel_ID'] = [ids for chunk in parcel_neighbour_chunks for ids in chunk[0]]
parcels['LINZ_parcel_sides_zones'] = [zones for chunk in parcel_neighbour_chunks for zones in chunk[1]]

j. i. LINZ_TRNSPWR_ohead_indicator Indicator (0 or 1) for LINZ parcel located under overhead transpower line. Leave blank otherwise. Note that ‘TRANSLINE’ denotes overhead transmission lines in the GIS dataset, while ‘CABLE’ denotes underground transmission lines and can be ignored.

j. ii. LINZ_TRNSPWR_ohead_name

Note: What name to use? 'descriptio'?

power = gpd.read_file('input/Transmission_Lines_exTRANSPOWER/Transmission_Lines.shp').to_crs(parcels.crs)
# only interested in overhead
power = power[power['type'] == 'TRANSLINE']
power.sample(3)

	OBJECTID	MXLOCATION	designvolt	status	descriptio	type	Symbol	Shape__Len	GlobalID	geometry
3	4	ALB-HPI-A	220	COMMISSIONED	Albany - Huapai A	TRANSLINE	220 TRANSLINE	14887.805068	8cb81a31-2c9c-497c-a8fc-646614b9290d	MULTILINESTRING ((1737714.820 5930696.980, 173...
82	84	HEN-MDN-A	220	COMMISSIONED	Henderson - Marsden A	TRANSLINE	220 TRANSLINE	120125.553279	ee8afe11-2e05-4727-9aca-aa8f88c2897d	MULTILINESTRING ((1744173.000 5921474.000, 174...
115	117	MDN-MPE-A	110	COMMISSIONED	Marsden - Maungatapere A	TRANSLINE	110 TRANSLINE	29289.440597	dc6f9edb-5717-4027-8fa5-26fa4f7c74b0	MULTILINESTRING ((1732389.000 6028862.000, 173...

%%time
# get dataframe associating parcel indices with overhead power lines
# alternative approach is to do how='left', then combine on index using group by, but that seems much slower when incorporating the results into the final gdf
power_intersect = gpd.sjoin(parcels, power[['descriptio', 'geometry']]).drop(columns=['index_right'])

power_intersect.index.value_counts()

377815    10
372782     6
338222     4
405352     4
393725     4
          ..
412524     1
463723     1
301926     1
11108      1
86014      1
Length: 5463, dtype: int64

%%time
def get_powerlines(id):
    if id in power_intersect.index:
        powerlines = power_intersect.loc[[id]]
        # filter out duplicates
        return powerlines.descriptio.unique().tolist()
#         return ','.join(powerlines.descriptio.unique())

    else:
        return None
parcel_powerlines = process_map(get_powerlines, list(parcels.index), max_workers=14, chunksize=1000)
parcels['LINZ_TRNSPWR_ohead_name'] = parcel_powerlines
parcels['LINZ_TRNSPWR_ohead_indicator'] = [int(p is not None) for p in parcel_powerlines]

CPU times: user 434 ms, sys: 16.2 ms, total: 450 ms
Wall time: 448 ms

ax = parcels[parcels['LINZ_TRNSPWR_ohead_indicator'] == 1].plot()
power.plot(column='designvolt', legend=True, ax=ax)
plt.xlim((1.72e6, 1.8e6))
plt.ylim((5.88e6, 6e6))
ctx.add_basemap(ax, source=ctx.providers.Esri.WorldImagery, crs=parcels.crs)

k. i. LINZ_VWSHFT_ohead_indicator Indicator (0 or 1) for LINZ parcel located under viewshafts. Leave blank otherwise.

k. ii. LINZ_VWSHFT_ohead_name Name of the volcanic cone (e.g. Mt Albert). Leave blank if no viewshaft applies.

k. iii. LINZ_VWSHFT_ohead_ID OBJECTID of the viewshaft. Leave blank if no viewshaft applies.

viewshafts_local = gpd.read_file('restricted/2016_Unitary_Plan_operational_15Nov/UP_OIP_15Nov2016_SHP/MASTER_UP_LocallySignificantVolcanicViewshafts.shp').to_crs(parcels.crs)
viewshafts_regional = gpd.read_file('restricted/2016_Unitary_Plan_operational_15Nov/UP_OIP_15Nov2016_SHP/MASTER_UP_RegionallySignificantVolcanicViewShaftsAndHeightSensitiveAreasOverlay.shp').to_crs(parcels.crs)
viewshafts_museum = gpd.read_file('restricted/2016_Unitary_Plan_operational_15Nov/UP_OIP_15Nov2016_SHP/MASTER_UP_AucklandMuseumViewshaftOverlay.shp').to_crs(parcels.crs)

# include dilworth?
viewshafts_dilworth = gpd.read_file('restricted/2016_Unitary_Plan_operational_15Nov/UP_OIP_15Nov2016_SHP/MASTER_UP_DilworthTerraceHousesViewshaftOverlay.shp').to_crs(parcels.crs)

viewshafts_museum['OBJECTID'] = ['Museum_' + str(s) for s in viewshafts_museum['OBJECTID']]
viewshafts_regional['OBJECTID'] = ['RSVS_' + str(s) for s in viewshafts_regional['OBJECTID']]
viewshafts_local['OBJECTID'] = ['LSVS_' + str(s) for s in viewshafts_local['OBJECTID']]

viewshafts = pd.concat([viewshafts_museum, viewshafts_local, viewshafts_regional])

%%time
joined = gpd.sjoin(parcels, viewshafts[["NAME", "OBJECTID", "geometry"]])

CPU times: user 7.3 s, sys: 211 ms, total: 7.51 s
Wall time: 7.51 s

%%time
def get_viewshafts(id):
    if id in joined.index:
        vs = joined.loc[[id]]
        # filter out duplicates
        return vs["OBJECTID"].unique().tolist(), vs["NAME"].unique().tolist()
    else:
        return None
    
parcel_viewshafts = process_map(get_viewshafts, list(parcels.index), max_workers=14, chunksize=1000)

parcels['LINZ_VWSHFT_ohead_name'] = [vs[1] if vs is not None else None for vs in parcel_viewshafts]
parcels['LINZ_VWSHFT_ohead_ID'] = [vs[0] if vs is not None else None for vs in parcel_viewshafts]
parcels['LINZ_VWSHFT_ohead_indicator'] = [int(p is not None) for p in parcel_viewshafts]

CPU times: user 3.02 s, sys: 3.51 s, total: 6.53 s
Wall time: 1min 41s

ax = parcels[parcels.LINZ_VWSHFT_ohead_indicator == 1].plot()
ctx.add_basemap(ax, source=ctx.providers.Esri.WorldImagery, crs=parcels.crs)

2. AUP shapefile information. Calculations are based on centroid of matched parcels:

a. LINZmatch_AUP_name AUP zone (string) consent is located in (from the AUP shapefiles)

b. LINZmatch_AUP_code AUP Zone Code (numerical) consent is located in

Done above in 1h and 1i.

c. Hdist_rural Minimum Haversine distance to nearest Rural Zone, plus

c. i. Hdist_rural_name Name of rural zone identified

c. ii. Hdist_rural_code Numeric code of rural zone identified

I've included all of these as rural:
['Rural - Mixed Rural Zone',
'Rural - Rural Coastal Zone',
'Rural - Countryside Living Zone',
'Rural - Rural Production Zone',
'Rural - Rural Conservation Zone',
'Rural - Waitakere Ranges Zone',
'Rural - Waitakere Foothills Zone']

%%time
aup_zones = gpd.read_file('restricted/2016_Unitary_Plan_operational_15Nov/UP_OIP_15Nov2016_SHP/MASTER_UP_BaseZone.shp')
aup_zones = aup_zones.to_crs(2193)
aup_zones.sample(3)

CPU times: user 38.9 s, sys: 3.85 s, total: 42.8 s
Wall time: 49.4 s

	WORKFLOWJO	last_edi00	CONTOUR	created_da	DocumentUR	GlobalID	GROUPZONE	GROUPZONE_	ID	NAME	OBJECTID	PARCEL_BAS	PARCEL_B00	PRECINCT	PRECINCT_r	PRECINCTGR	PRECINCT00	SCHEDULE	SHAPE_Area	SHAPE_Leng	STATUS	SUBPRECINC	SUBPRECI00	SUBTYPE	SUBTYPE_re	TYPE	TYPE_resol	VALIDATION	VALIDATI00	VERSIONSTA	VERSIONS00	ZONE	ZONE_resol	ZONEHEIGHT	geometry
102760	NaN	20161111012	None	20160718211	None	{D2CB42B3-4AC0-494F-B7DB-1B625F85309E}	7	General	None	None	102761	None	None	None	None	None	None	None	4820.266107	506.146565	None	None	None	None	None	None	None	3	Valid and Public	4	Operative	27	Road	NaN	POLYGON ((1757938.586 5918711.839, 1757928.533...
55389	NaN	20161111011	None	20160718211	None	{7EE47EE9-4168-4027-B43B-AD167FAA578E}	2	Residential	None	None	55390	None	None	None	None	None	None	None	12835.318390	716.757720	None	None	None	None	None	None	None	3	Valid and Public	4	Operative	18	Residential - Mixed Housing Suburban Zone	NaN	POLYGON ((1768405.622 5906733.976, 1768409.836...
64035	NaN	20161111011	None	20160718211	None	{6204C0B4-9FC2-4450-9C81-1233B9CCD795}	5	Coastal	None	None	64036	None	None	None	None	None	None	None	10.307124	25.986136	None	None	None	None	None	None	None	3	Valid and Public	4	Operative	30	Coastal - General Coastal Marine Zone	NaN	POLYGON ((1714074.272 5961757.993, 1714076.965...

rural_codes = aup_zones[aup_zones.ZONE_resol.str.lower().str.contains('rural - ', na=False)]['ZONE'].unique()

# check that each rural zone code matches with a unique rural zone name 
assert all([len(aup_zones[aup_zones.ZONE == code].ZONE_resol.unique()) == 1 for code in rural_codes])
# dictionary mapping code to names
rural_code2name = {code: aup_zones[aup_zones.ZONE == code].ZONE_resol.unique()[0] for code in rural_codes}

aup_zones[aup_zones.ZONE_resol.isna()]

	WORKFLOWJO	last_edi00	CONTOUR	created_da	DocumentUR	GlobalID	GROUPZONE	GROUPZONE_	ID	NAME	OBJECTID	PARCEL_BAS	PARCEL_B00	PRECINCT	PRECINCT_r	PRECINCTGR	PRECINCT00	SCHEDULE	SHAPE_Area	SHAPE_Leng	STATUS	SUBPRECINC	SUBPRECI00	SUBTYPE	SUBTYPE_re	TYPE	TYPE_resol	VALIDATION	VALIDATI00	VERSIONSTA	VERSIONS00	ZONE	ZONE_resol	ZONEHEIGHT	geometry
20853	NaN	20161111011	None	20160718211	None	{2D83F680-9587-4D57-AF0B-CB7CA27C3D2C}	6	Special purpose zone	None	None	20854	None	None	None	None	None	None	None	22.286945	72.629617	None	None	None	None	None	None	None	3	Valid and Public	4	Operative	58	None	NaN	POLYGON Z ((1767684.860 5903714.023 33.721, 17...
121765	NaN	20161111010	None	20160718211	None	{FD3E8BB4-1979-42B5-9A77-5D04AE5190AF}	6	Special purpose zone	None	None	121766	None	None	None	None	None	None	None	4.114962	86.135341	None	None	None	None	None	None	None	3	Valid and Public	4	Operative	58	None	NaN	POLYGON Z ((1767684.083 5903713.690 33.721, 17...

# 2 NAs in ZONE_resol are from a zone 58, which only has observations
aup_zones[aup_zones.ZONE == '58']

	WORKFLOWJO	last_edi00	CONTOUR	created_da	DocumentUR	GlobalID	GROUPZONE	GROUPZONE_	ID	NAME	OBJECTID	PARCEL_BAS	PARCEL_B00	PRECINCT	PRECINCT_r	PRECINCTGR	PRECINCT00	SCHEDULE	SHAPE_Area	SHAPE_Leng	STATUS	SUBPRECINC	SUBPRECI00	SUBTYPE	SUBTYPE_re	TYPE	TYPE_resol	VALIDATION	VALIDATI00	VERSIONSTA	VERSIONS00	ZONE	ZONE_resol	ZONEHEIGHT	geometry
20853	NaN	20161111011	None	20160718211	None	{2D83F680-9587-4D57-AF0B-CB7CA27C3D2C}	6	Special purpose zone	None	None	20854	None	None	None	None	None	None	None	22.286945	72.629617	None	None	None	None	None	None	None	3	Valid and Public	4	Operative	58	None	NaN	POLYGON Z ((1767684.860 5903714.023 33.721, 17...
121765	NaN	20161111010	None	20160718211	None	{FD3E8BB4-1979-42B5-9A77-5D04AE5190AF}	6	Special purpose zone	None	None	121766	None	None	None	None	None	None	None	4.114962	86.135341	None	None	None	None	None	None	None	3	Valid and Public	4	Operative	58	None	NaN	POLYGON Z ((1767684.083 5903713.690 33.721, 17...

rural = aup_zones[aup_zones.ZONE.isin(rural_codes)]

rural_by_zone_dict = {code: rural[rural.ZONE == code].dissolve() for code in rural_codes}

def find_nearest(item):
    index, row = item
    distance_candidates = []
    code_candidates = []
    for code, rural_gdf in rural_by_zone_dict.items():
        distance_candidates.append(row.geometry.distance(rural_gdf.geometry[0]))
        code_candidates.append(rural_gdf.ZONE[0])
    return distance_candidates, code_candidates

# this might hang for a few minutes before multiprocessing starts
output = process_map(find_nearest, parcels_sample.iterrows(), max_workers=14, chunksize=1, total=len(parcels_sample))

# all distances (to any zone)
distance_candidates = np.array([x[0] for x in output])
code_candidates = np.array([x[1] for x in output])

# indices of minimum distances
min_idx = np.argmin(distance_candidates, axis=-1)

parcels_sample['Hdist_rural'] = distance_candidates[(np.arange(len(distance_candidates)), min_idx)]
parcels_sample['Hdist_rural_code'] = code_candidates[(np.arange(len(distance_candidates)), min_idx)]
parcels_sample['Hdist_rural_name'] = parcels_sample.apply(lambda x: rural_code2name[x.Hdist_rural_code], axis=1)

import matplotlib.patches as mpatches
import matplotlib.pyplot as plt

colours = ('blue', 'orange', 'green', 'red', 'purple', 'brown', 'pink', 'cyan')
name2colour = {name: colour for name, colour in zip(rural_code2name.values(), colours)}

column = 'Hdist_rural'
subsample = parcels_sample[parcels_sample[column] > 0].sample(5)
subsample['buffer'] = subsample.apply(lambda x: x.geometry.buffer(x[column]), axis=1)
subsample['geometry'] = subsample['buffer']
subsample = subsample[~subsample.is_empty]
# ax = rural.plot(column='ZONE_resol', legend=True)
# subsample.plot(column='Hdist_rural_name', alpha=0.4, ax=ax)
ax = rural.plot(color=[name2colour[z] for z in rural.ZONE_resol])
subsample.plot(color=[name2colour[z] for z in subsample.Hdist_rural_name], alpha=0.65, ax=ax)
# plt.ylim((5.88e6, 5.96e6))

# add legend
legend_patches = [mpatches.Patch(color=col, label=name) for name, col in name2colour.items()]
plt.legend(handles=legend_patches)

# ctx.add_basemap(ax=ax, crs=subsample.crs)

<matplotlib.legend.Legend at 0x7f0d8c1423d0>

d. Hdist_bus Minimum Haversine distance to nearest Business Zone

d. i. Hdist_bus_name

d. ii. Hdist_bus_code

business_codes = aup_zones[aup_zones.ZONE_resol.str.lower().str.contains('business - ', na=False)]['ZONE'].unique()

# check that each business zone code matches with a unique business zone name 
assert all([len(aup_zones[aup_zones.ZONE == code].ZONE_resol.unique()) == 1 for code in business_codes])
# dictionary mapping code to names
business_code2name = {code: aup_zones[aup_zones.ZONE == code].ZONE_resol.unique()[0] for code in business_codes}
business_code2name

{'44': 'Business - Neighbourhood Centre Zone',
 '12': 'Business - Mixed Use Zone',
 '17': 'Business - Light Industry Zone',
 '5': 'Business - Heavy Industry Zone',
 '49': 'Business - General Business Zone',
 '1': 'Business - Business Park Zone',
 '22': 'Business - Town Centre Zone',
 '10': 'Business - Metropolitan Centre Zone',
 '7': 'Business - Local Centre Zone',
 '35': 'Business - City Centre Zone'}

business = aup_zones[aup_zones.ZONE.isin(business_codes)]

business_by_zone_dict = {code: business[business.ZONE == code].dissolve() for code in business_codes}

def find_nearest(item):
    index, row = item
    distance_candidates = []
    code_candidates = []
    for code, business_gdf in business_by_zone_dict.items():
        distance_candidates.append(row.geometry.distance(business_gdf.geometry[0]))
        code_candidates.append(business_gdf.ZONE[0])
    return distance_candidates, code_candidates

# this might hang for a few minutes before multiprocessing starts
output = process_map(find_nearest, parcels_sample.iterrows(), max_workers=14, chunksize=1, total=len(parcels_sample))

# all distances (to any zone)
distance_candidates = np.array([x[0] for x in output])
code_candidates = np.array([x[1] for x in output])

# indices of minimum distances
min_idx = np.argmin(distance_candidates, axis=-1)

parcels_sample['Hdist_bus'] = distance_candidates[(np.arange(len(distance_candidates)), min_idx)]
parcels_sample['Hdist_bus_code'] = code_candidates[(np.arange(len(distance_candidates)), min_idx)]
parcels_sample['Hdist_bus_name'] = parcels_sample.apply(lambda x: business_code2name[x.Hdist_bus_code], axis=1)

import matplotlib.patches as mpatches
import matplotlib.pyplot as plt

colours = ('blue', 'orange', 'green', 'red', 'purple', 'brown', 'pink', 'cyan', 'black', 'white')
name2colour = {name: colour for name, colour in zip(business_code2name.values(), colours)}

# hard to see, subset to smaller area
bounds = {'x1': 1.76e6, 'x2': 1.7613e6, 'y1': 5.9123e6, 'y2': 5.914e6}

column = 'Hdist_bus'
subsample = parcels_sample.cx[bounds['x1']:bounds['x2'], bounds['y1']:bounds['y2']]
subsample['buffer'] = subsample.apply(lambda x: x.geometry.buffer(x[column]), axis=1)
subsample['geometry'] = subsample['buffer']

subsample = subsample[~subsample.is_empty]

business_plot = business.cx[bounds['x1']:bounds['x2'], bounds['y1']:bounds['y2']]
ax = business_plot.plot(color=[name2colour[z] for z in business_plot.ZONE_resol])
subsample.plot(color=[name2colour[z] for z in subsample.Hdist_bus_name], alpha=0.65, ax=ax)
# plt.ylim((5.88e6, 5.96e6))

# add legend
legend_patches = [mpatches.Patch(color=col, label=name) for name, col in name2colour.items()]
plt.legend(handles=legend_patches)

ctx.add_basemap(ax=ax, crs=subsample.crs)

e. Hdist_resid Minimum Haversine distance to nearest Residential Zone,

e. i. Hdist_resid_name

e. ii. Hdist_resid_code

resid_codes = aup_zones[aup_zones.ZONE_resol.str.lower().str.contains('resid', na=False)]['ZONE'].unique()

resid_codes

array(['60', '23', '18', '8', '19', '20'], dtype=object)

# check that each resid zone code matches with a unique resid zone name 
assert all([len(aup_zones[aup_zones.ZONE == code].ZONE_resol.unique()) == 1 for code in resid_codes])
# dictionary mapping code to names
resid_code2name = {code: aup_zones[aup_zones.ZONE == code].ZONE_resol.unique()[0] for code in resid_codes}
resid_code2name

{'60': 'Residential - Mixed Housing Urban Zone',
 '23': 'Residential - Large Lot Zone',
 '18': 'Residential - Mixed Housing Suburban Zone',
 '8': 'Residential - Terrace Housing and Apartment Building Zone',
 '19': 'Residential - Single House Zone',
 '20': 'Residential - Rural and Coastal Settlement Zone'}

resid = aup_zones[aup_zones.ZONE.isin(resid_codes)]

resid_by_zone_dict = {code: resid[resid.ZONE == code].dissolve() for code in resid_codes}

def find_nearest(item):
    index, row = item
    distance_candidates = []
    code_candidates = []
    for code, gdf in resid_by_zone_dict.items():
        distance_candidates.append(row.geometry.distance(gdf.geometry[0]))
        code_candidates.append(gdf.ZONE[0])
    return distance_candidates, code_candidates

# this might hang for a few minutes before multiprocessing starts
output = process_map(find_nearest, parcels_sample.iterrows(), max_workers=14, chunksize=1, total=len(parcels_sample))

# all distances (to any zone)
distance_candidates = np.array([x[0] for x in output])
code_candidates = np.array([x[1] for x in output])

# indices of minimum distances
min_idx = np.argmin(distance_candidates, axis=-1)

parcels_sample['Hdist_resid'] = distance_candidates[(np.arange(len(distance_candidates)), min_idx)]
parcels_sample['Hdist_resid_code'] = code_candidates[(np.arange(len(distance_candidates)), min_idx)]
parcels_sample['Hdist_resid_name'] = parcels_sample.apply(lambda x: resid_code2name[x.Hdist_resid_code], axis=1)

import matplotlib.patches as mpatches
import matplotlib.pyplot as plt

colours = ('blue', 'orange', 'green', 'red', 'purple', 'brown', 'pink', 'cyan', 'black', 'white')
name2colour = {name: colour for name, colour in zip(resid_code2name.values(), colours)}

# hard to see, subset to smaller area
bounds = {'x1': 1.76e6, 'x2': 1.7613e6, 'y1': 5.9123e6, 'y2': 5.914e6}

column = 'Hdist_resid'
subsample = parcels_sample.cx[bounds['x1']:bounds['x2'], bounds['y1']:bounds['y2']]
subsample['buffer'] = subsample.apply(lambda x: x.geometry.buffer(x[column]), axis=1)
subsample['geometry'] = subsample['buffer']

subsample = subsample[~subsample.is_empty]

resid_plot = resid.cx[bounds['x1']:bounds['x2'], bounds['y1']:bounds['y2']]
ax = resid_plot.plot(color=[name2colour[z] for z in resid_plot.ZONE_resol])
subsample.plot(color=[name2colour[z] for z in subsample.Hdist_resid_name], alpha=0.65, ax=ax)
# plt.ylim((5.88e6, 5.96e6))

# add legend
legend_patches = [mpatches.Patch(color=col, label=name) for name, col in name2colour.items()]
plt.legend(handles=legend_patches)

ctx.add_basemap(ax=ax, crs=subsample.crs)

/data/miniconda3/envs/house-upzone/lib/python3.8/site-packages/geopandas/geodataframe.py:1322: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  super(GeoDataFrame, self).__setitem__(key, value)

f. Hdist_SH Minimum Haversine distance to Residential - Single House Zone

g. Hdist_MHS Minimum Haversine distance to Residential - Mixed Housing Suburban Zone

h. Hdist_MHU Minimum Haversine distance to Residential - Mixed Housing Urban Zone

i. Zone Hdist_THA Minimum Haversine distance to Residential - Terrace Housing and Apartments

Note: this is the real name for i: 'Residential - Terrace Housing and Apartment Building Zone'

%%time
postfix2name = {
            'SH': 'Residential - Single House Zone',
            'MHS': 'Residential - Mixed Housing Suburban Zone',
            'MHU': 'Residential - Mixed Housing Urban Zone',
            'THA': 'Residential - Terrace Housing and Apartment Building Zone'
               }
for postfix, zone in tqdm(postfix2name.items()):
    resid_gdf = resid[resid.ZONE_resol == zone].dissolve()
    def get_distance(geo):
        return geo.distance(resid_gdf.geometry[0])
    parcels_sample[f'Hdist_{postfix}'] = process_map(get_distance, parcels_sample.geometry, max_workers=14, chunksize=1)

CPU times: user 48.5 s, sys: 14.3 s, total: 1min 2s
Wall time: 1min 43s

postfix = 'THA'
column = f'Hdist_{postfix}'
subsample = parcels_sample.sample(10)
subsample['buffer'] = subsample.apply(lambda x: x.geometry.buffer(x[column]), axis=1)
subsample['geometry'] = subsample['buffer']
ax = subsample.plot(color='red', alpha=0.4)
resid[resid.ZONE_resol == postfix2name[postfix]].plot(ax=ax)
ctx.add_basemap(ax, crs=2193)

3. Local Area (LA) Information (information on the LA that the consent is located in):

a. Name (just the string is fine; I did not generate numeric codes for the local areas generated) Local_Area_name

LA = gpd.read_file('input/Modified_Community_Boards_SHP.zip').to_crs(2193)
LA.sample(3)

	OBJECTID	Local_Area	geometry
8	10.0	Hibiscus Coast	MULTIPOLYGON (((1752023.352 5954803.281, 17520...
17	21.0	Papatoetoe	POLYGON ((1765105.421 5908611.893, 1765113.661...
2	25.0	Rodney-Kumeu-Riverhead	POLYGON ((1742532.808 5931237.574, 1742490.377...

parcels_sample['geometry'] = parcels_sample['geometry_centroid_2193']
parcels_sample = gpd.sjoin(parcels_sample, LA[['Local_Area', 'geometry']]).drop(columns=['index_right'])
parcels_sample = parcels_sample.rename(columns={'Local_Area': 'Local_Area_name'})

4. 2018 Statistical Area 2 Information (information on the 2018SA2 that the consent is located in)

a. Name (string) SA22018_name

b. Numeric code SA22018_code

sa2 = gpd.read_file('NZ-SA/statistical-area-2-2020-generalised.gdb').to_crs(2193)
sa2.sample(3)

%%time
parcels_sample['geometry'] = parcels_sample['geometry_centroid_2193']
parcels_sample = gpd.sjoin(parcels_sample, sa2[['SA22020_V1_00_NAME', 'SA22020_V1_00', 'geometry']]).drop(columns=['index_right'])
parcels_sample = parcels_sample.rename(columns={'SA22020_V1_00_NAME': 'SA22018_name', 'SA22020_V1_00': 'SA22018_code'})

CPU times: user 8.29 s, sys: 3.88 ms, total: 8.29 s
Wall time: 8.29 s

5. Area Unit Information (information on the 2013AU  that the consent is located in):

a. Name (string) AU2013_name

b. Numeric code AU2013_code

au2013 = gpd.read_file('input/area-unit-2013.gdb.zip').to_crs(2193)
au2013.sample(3)

	AU2013_V1_00	AU2013_V1_00_NAME	AREA_SQ_KM	LAND_AREA_SQ_KM	Shape_Length	geometry
1481	573903	Newlands North	0.804087	0.804087	5958.640337	MULTIPOLYGON (((174.82896 -41.21949, 174.83011...
958	525202	Kawakawa-Orere	105.216001	105.216001	57437.556145	MULTIPOLYGON (((175.14259 -36.93214, 175.14265...
608	597102	Inland Water-Lake Ellesmere South	63.304671	0.000000	75233.330853	MULTIPOLYGON (((172.57398 -43.76779, 172.57401...

%%time
parcels_sample['geometry'] = parcels_sample['geometry_centroid_2193']
parcels_sample = gpd.sjoin(parcels_sample, au2013[['AU2013_V1_00_NAME', 'AU2013_V1_00', 'geometry']]).drop(columns=['index_right'])
parcels_sample = parcels_sample.rename(columns={'AU2013_V1_00_NAME': 'AU2013_name', 'AU2013_V1_00': 'AU2013_code'})

CPU times: user 6.96 s, sys: 0 ns, total: 6.96 s
Wall time: 6.96 s

6. 2018 Meshblock Information (information on the 2018MB  that the consent is located in):

a. Code MB2018_code

mb2018 = gpd.read_file('input/meshblock-2018-clipped-generalised.gdb.zip').to_crs(4326)
mb2018.sample(3)

	MB2018_V1_00	LANDWATER	LANDWATER_NAME	LAND_AREA_SQ_KM	AREA_SQ_KM	SHAPE_Length	geometry
52865	4011926	12	Mainland	0.173710	0.173710	2129.422679	MULTIPOLYGON (((174.91945 -36.94519, 174.92053...
23832	1429800	12	Mainland	0.021918	0.021918	731.394363	MULTIPOLYGON (((176.91682 -39.49291, 176.91702...
15554	0759520	12	Mainland	0.020695	0.020695	928.355735	MULTIPOLYGON (((174.91840 -37.01411, 174.91871...

%%time
parcels_sample['geometry'] = parcels_sample['geometry_centroid']
parcels_sample = gpd.sjoin(parcels_sample, mb2018[['MB2018_V1_00', 'geometry']]).drop(columns=['index_right'])
parcels_sample = parcels_sample.rename(columns={'MB2018_V1_00': 'MB2018_code'})

CPU times: user 11.3 s, sys: 8.2 ms, total: 11.3 s
Wall time: 11.3 s

7. 2013 Meshblock Information (information on the 2013MB  that the consent is located in):

a. Code MB2013_code

mb2013 = gpd.read_file('input/meshblock-2013.gdb.zip').to_crs(4326)
mb2013.sample(3)

%%time
parcels_sample['geometry'] = parcels_sample['geometry_centroid_2193']
parcels_sample = gpd.sjoin(parcels_sample, mb2013[['MeshblockNumber', 'geometry']]).drop(columns=['index_right'])
parcels_sample = parcels_sample.rename(columns={'MeshblockNumber': 'MB2013_code'})

CPU times: user 11.4 s, sys: 7.05 ms, total: 11.4 s
Wall time: 11.4 s

8. Additional distance information from consent location

For these distance calculations, use EPSG 2193 (less distortion).

parcels_sample['geometry'] = parcels_sample['geometry_centroid']
parcels_sample = parcels_sample.to_crs(2193)
parcels_sample.sample(3)

	id	appellation	affected_surveys	parcel_intent	topology_type	statutory_actions	land_district	titles	survey_area	calc_area	geometry	geometry_centroid	geometry_polygon	SA22018_name	SA22018_code	MB2013_code	MB2018_code	AU2013_name	AU2013_code
102782	4896031	Lot 1 DP 61837	DP 61837	DCDB	Primary	None	North Auckland	NA17C/935	685.0	685.0	POINT (1773608.374 5914596.096)	POINT (174.94849 -36.89867)	MULTIPOLYGON (((174.94827 -36.89874, 174.94841...	Cockle Bay	153400	0655102	0655102	Cockle Bay	521502
119786	4931702	Lot 4 DP 166054	DP 166054	DCDB	Primary	None	North Auckland	NA100D/34	52130.0	50575.0	POINT (1783027.533 5889659.615)	POINT (175.06020 -37.12153)	MULTIPOLYGON (((175.06295 -37.12105, 175.06293...	Ararimu	166400	0815302	0815302	Hunua	521132
362342	7415715	Lot 23 DP 455616	DP 455616, DP 462513	Fee Simple Title	Primary	None	North Auckland	586776	103.0	103.0	POINT (1763989.858 5916157.717)	POINT (174.84025 -36.88632)	MULTIPOLYGON (((174.84016 -36.88629, 174.84020...	Stonefields West	144900	0465305	4000293	Stonefields	517202

a. Minimum Haversine distance to coastline Hdist_coast

There are a few different datasets that could be used for this:

- NZ Coastlines (Topo 1:50k) https://data.linz.govt.nz/layer/50258-nz-coastlines-topo-150k/
- NZ Coastline - mean high water https://data.linz.govt.nz/layer/105085-nz-coastline-mean-high-water/
- NZ Coastlines and Islands Polygons (Topo 1:50k) https://data.linz.govt.nz/layer/51153-nz-coastlines-and-islands-polygons-topo-150k/  

The first doesn't have islands (e.g. Waiheke).
The second is probably most appropriate.

%%time
coastline = gpd.read_file('input/lds-nz-coastline-mean-high-water-FGDB.zip!nz-coastline-mean-high-water.gdb').to_crs(2193)
coastline = coastline.cx[bounds2193['x1']:bounds2193['x2'], bounds2193['y1']:bounds2193['y2']]
coastline_dissolved = coastline.dissolve()

%%time
parcels_sample['Hdist_coast'] = parcels_sample.apply(lambda x: x.geometry.distance(coastline_dissolved.geometry[0]), axis=1)

CPU times: user 14.3 s, sys: 0 ns, total: 14.3 s
Wall time: 14.3 s

# if distance work, then red circles should extend to the nearest coastline, and no further
subsample = parcels_sample.sample(10)
subsample['coast_buffer'] = subsample.apply(lambda x: x.geometry.buffer(x.Hdist_coast), axis=1)
subsample['geometry'] = subsample['coast_buffer']
ax = subsample.plot(color='red', alpha=0.4)
coastline.cx[1.7e6:1.8e6, 5.85e6:5.97e6].plot(ax=ax)

<AxesSubplot:>

b. Minimum Haversinedistance to Open Space(set of AUP Zones) Hdist_open

c. Minimum Haversine distance to motorway Hdist_motorway

d. Minimum Haversine distance to main arterial road Hdist_main_road

parcels_sample['geometry'] = parcels_sample['geometry_centroid_2193']
parcels_sample = parcels_sample.set_crs(2193)

roads = gpd.read_file('input/lds-nz-road-centrelines-topo-150k-FGDB.zip!nz-road-centrelines-topo-150k.gdb').to_crs(2193)
roads = roads.cx[bounds2193['x1']:bounds2193['x2'], bounds2193['y1']:bounds2193['y2']]
highways = roads[~roads.hway_num.isna()]
highways_dissolved = highways.dissolve()
arterial_roads = gpd.read_file('restricted/2016_Unitary_Plan_operational_15Nov/UP_OIP_15Nov2016_SHP/MASTER_UP_ArterialRoad.shp').to_crs(2193)
arterial_roads_dissolved = arterial_roads.dissolve()

ax = highways.plot()
arterial_roads.plot(ax=ax, color='red')
ctx.add_basemap(ax, crs=2193)

parcels_sample['Hdist_motorway'] = parcels_sample.progress_apply(lambda x: x.geometry.distance(highways_dissolved.geometry[0]), axis=1)
parcels_sample['Hdist_main_road'] = parcels_sample.progress_apply(lambda x: x.geometry.distance(arterial_roads_dissolved.geometry[0]), axis=1)

# if distance work, then red circles should extend to the nearest coastline, and no further
subsample = parcels_sample.sample(10)
subsample['highway_buffer'] = subsample.apply(lambda x: x.geometry.buffer(x.Hdist_motorway), axis=1)
subsample['geometry'] = subsample['highway_buffer']
ax = subsample.plot(color='red', alpha=0.4)
highways.plot(ax=ax)

<AxesSubplot:>

e. Minimum Haversine distance to rail line Hdist_rail

railroads = gpd.read_file('input/lds-nz-railway-centrelines-topo-150k-SHP.zip').to_crs(2193)
railroads = railroads.cx[bounds2193['x1']:bounds2193['x2'], bounds2193['y1']:bounds2193['y2']]
railroads_dissolved = railroads.dissolve()

parcels_sample['geometry'] = parcels_sample['geometry_centroid_2193']
parcels_sample['Hdist_rail'] = parcels_sample.apply(lambda x: x.geometry.distance(railroads_dissolved.geometry[0]), axis=1)

# if distance work, then red circles should extend to the nearest coastline, and no further
subsample = parcels_sample.sample(10)
subsample['rail_buffer'] = subsample.apply(lambda x: x.geometry.buffer(x.Hdist_rail), axis=1)
subsample['geometry'] = subsample['rail_buffer']
ax = subsample.plot(color='red', alpha=0.4)
railroads_dissolved.plot(ax=ax)
ctx.add_basemap(ax, crs=2193)

f. Haversine distance to downtown (use Skytower coordinates) Hdist_skytower

skytower = [-36.84838748948485, 174.7621736911587]
skytower = gpd.points_from_xy(x=[skytower[1]], y=[skytower[0]])
skytower = gpd.GeoDataFrame([{"name": "Skytower", "value": 1}], geometry=skytower, crs="EPSG:4326").to_crs(epsg=2193)

parcels_sample['geometry'] = parcels_sample['geometry_centroid_2193']
parcels_sample['Hdist_skytower'] = parcels_sample.apply(lambda x: x.geometry.distance(skytower.geometry[0]), axis=1)

# if distance works, then red circles should extend to the nearest sky tower, and no further
subsample = parcels_sample.sample(10)
subsample['skytower_buffer'] = subsample.apply(lambda x: x.geometry.buffer(x.Hdist_skytower), axis=1)
subsample['geometry'] = subsample['skytower_buffer']
ax = subsample.plot(color='red', alpha=0.2)
skytower.plot(ax=ax, color='black')
coastline.cx[1.7e6:1.8e6, 5.85e6:5.97e6].plot(ax=ax)

<AxesSubplot:>

9. Special Housing Area (SpHAs)

Indicator (1 or 0) for consent located in SpHAs SpHA_indicator
Note: here I've used centroids. Maybe should use parcel polygons instead.

spha = gpd.read_file('input/AC_Special_Housing_Area.zip').to_crs(2193)
spha_dissolved = spha.dissolve()
assert(len(spha_dissolved) == 1)

%%time
parcels_sample['geometry'] = parcels_sample['geometry_centroid_2193']
parcels_sample['SpHA_indicator'] = parcels_sample.apply(lambda x: spha_dissolved.geometry.contains(x.geometry), axis=1)

# if distance works, then red circles should extend to the nearest sky tower, and no further
subsample = parcels_sample.sample(500)
ax=subsample.plot(column='SpHA_indicator', alpha=0.6)
plt.ylim((5.89e6, 5.95e6))
plt.xlim((1.73e6, 1.78e6))
spha_dissolved.boundary.plot(ax=ax)
ctx.add_basemap(ax, crs=spha_dissolved.crs)

10. Ranged Address Indicator

11. Matching Type Indicators

a. LINZ_MATCH_CODE

b. LINZ_2ND_MATCH_CODE